ORIGINAL INVESTIGATION

Memory and psychostimulants: modulation of Pavlovianfear conditioning by amphetamine in C57BL/6 mice

Suzanne C. Wood & Stephan G. Anagnostaras

Received: 15 April 2007 /Accepted: 22 April 2008 / Published online: 15 May 2008# Springer-Verlag 2008

AbstractRationale and objectives With the use of prescriptionstimulants on the rise, it is important to examine thecognitive effects of low and moderate doses of stimulantsrather than only those typical of addicts.Materials and methods The present study examined theeffects a range of doses (0.005–8 mg/kg) of D-amphetaminesulfate on cued and contextual Pavlovian fear conditioningin mice.Results In agreement with previous research, subjectsadministered with a moderately high dose of amphetamine(8 mg/kg) pre-training, typical of what addicts might take,displayed impaired conditioned freezing when tested off-drug. Alternately, subjects injected with a very low dose ofamphetamine (0.005, 0.025, or 0.05 mg/kg) pre-training,similar to the therapeutic doses for attention deficithyperactivity disorder, displayed enhanced memory whentested off-drug. A control study showed that these effectswere not due to state-dependent learning.Conclusions Thus, dose is a critical determinant of thecognitive effects of psychostimulants.

Keywords Amphetamine . Fear conditioning .

Learning andmemory .Mouse

Introduction

Amphetamine can be highly addictive, leading individuals toexhibit behaviors ranging from relatively minor cognitiveimpairments to severe psychotic symptoms. Methamphet-amine abusers display worse performance on tests of wordrecall, perceptual speed, vocabulary, and abstract reasoning,compared with controls (Simon et al. 2002). After prolongedamphetamine abuse, addicts may also exhibit symptomssimilar to schizophrenia, commonly referred to as amphet-amine or stimulant psychosis (Harris and Batki 2000).

Conversely, amphetamine is also used to treat ailmentssuch as narcolepsy and attention deficit hyperactivitydisorder (ADHD). Amphetamine (e.g., Adderall®)improves ADHD symptoms for most affected children(Ahmann et al. 2001), and several related stimulants aresimilarly effective (e.g., methamphetamine, methylpheni-date, atomoxetine, modafinil; Leonard et al. 2004).

While prescription stimulants containing amphetaminehave proven beneficial to many people suffering fromADHD, excessive daytime sleepiness, and narcolepsy, off-label use of these drugs is a growing problem. The life-timeprevalence of non-medical use of prescription stimulants incollege students has been recently reported to be 6.9%(McCabe et al. 2005) to 8.3% (Teter et al. 2006). Thesedrugs tend to be regarded by the public as cognitiveenhancers, presumably by promoting mental arousal orwakefulness. “Academic doping” is an emerging phenom-enon in many educational settings (Butcher 2003). Stimu-lants also have documented benefits (and problems) inmilitary use (Caldwell et al. 1995; Cornum et al. 1995;Cornum 1994). Indeed, a growing body of evidencedemonstrates that healthy volunteers, not just those withADHD, display cognitive benefits from low doses ofstimulants (Barch and Carter 2005; Turner et al. 2003).

Psychopharmacology (2009) 202:197–206DOI 10.1007/s00213-008-1185-9

S. C. Wood (*) : S. G. AnagnostarasMolecular Cognition Laboratory, Department of Psychology,University of California, San Diego,9500 Gilman Dr. MC 0109, La Jolla,San Diego, CA 92093-0109, USAe-mail: scwood@ucsd.edu

S. G. AnagnostarasProgram in Neurosciences, University of California, San Diego,9500 Gilman Dr. MC 0109, La Jolla,San Diego, CA 92093-0109, USA

These trends illuminate the importance of studying theeffects of a variety of acute doses of amphetamine ratherthan focusing solely on the higher, chronic doses typical ofaddicts. The current study examines the effects of a rangeof acute doses of D-amphetamine on a standard rodentlearning and memory task, Pavlovian fear conditioning.

In Pavlovian fear conditioning, a discrete neutral stimulus(e.g., a tone) is paired with an aversive stimulus (e.g., afootshock). After training, when presented with the discretestimulus, alone, the subject exhibits fear. In addition, uponreturning to the environmental chamber in which it had beentrained, the subject also exhibits fear, a phenomenon knownas context conditioning. A common measure of conditionedfear in rodents is freezing, or the absence of all movements,excluding respiration (Fanselow 1980).

The neurobiology of Pavlovian fear conditioning hasbeen studied extensively. The dorsal hippocampus iscritically involved in encoding memory of the context in atime-graded manner (Anagnostaras et al. 1999, 2001).Because of the efficiency of contextual fear conditioning,it has become a leading model of explicit memory in ratsand mice (Anagnostaras et al. 2000, 2001). The amygdala,and specifically the basolateral/lateral complex of theamygdala, is involved in encoding an aversive associationwith both the context and tone (Fanselow and Gale 2003;Fanselow and Poulos 2005). Amphetamine has been shownto alter amygdalar activity in a number of ways, includingpotentiating the synaptic transmission between the amyg-dala and nucleus accumbens (Kessal et al. 2005). Thehippocampus has also been implicated in amphetamine-induced locomotor activity as lesions of the ventralhippocampus disrupt amphetamine-induced locomotion,and stimulation enhances it (White et al. 2006).

We previously found (Wood et al. 2007) that an acutedose of cocaine enhances learning on Pavlovian fearconditioning when given at a very low dose (0.1 mg/kg).Interestingly, we also found that a moderate dose of cocaine(15 mg/kg) disrupted conditioned freezing. These effectswere general to both contextual and cued fear. These datareinforce the human findings that dosage is the criticaldeterminant of whether a particular stimulant enhances orimpairs memory. In particular, doses similar to those givenfor ADHD often improve attention, academic performance,and reduce impulsivity in humans. In contrast, high dosessuch as those taken by addicts are associated withunemployment, reduced executive function, anxiety, psy-chosis, and of course, addiction (Ellinwood et al. 1998).One potential confound of our prior study is that cocaineproduces local anesthesia, which could directly reduce fearconditioning by reducing the painfulness of the footshock.Therefore, in the present study, we examined the effects ofamphetamine on conditioned fear in order to extend thefindings to a widely prescribed stimulant, as well as to

control for local anesthesia. We also conducted a controlstudy to address the issue of state-dependent effects onlearning. We predicted that amphetamine would show thesame pattern as cocaine, enhancing Pavlovian fear condition-ing at low doses and interfering with fear conditioning athigher doses. The results of the current study are consistentwith this hypothesis, showing enhanced learning duringacquisition and recall of cued fear at low doses of amphet-amine (0.005, 0.025, and 0.05 mg/kg) and deficits inconditioned freezing at higher doses (4 and 8 mg/kg). Inaddition, there was no evidence found to support the idea thatstate-dependent learning was a factor in the present results.

Materials and methods

Subjects

One-hundred thirty-six (68 male, 68 female) C57BL6/NCrl(B6; Charles River Laboratories, San Diego, CA, USA)mice were used for experiment 1, and 40 (19 male, 21female) B6129SF1/J (H; Jackson Laboratory, West Sacra-mento, CA, USA) hybrid mice were used for experiment 2.Mice were weaned at 3 weeks of age and were group-housed (two to five mice per same sex cage), withcontinuous access to food and water. The animal colonywas maintained on a 14:10 light/dark schedule, and all testswere performed during the light phase of the cycle. Micewere at least 8 weeks old before testing. All animal care andtesting procedures were approved by the UCSD IACUCand were in accordance with the NIH “Principles oflaboratory animal care.” B6 and H mice were chosenbecause they display robust conditioned freezing and typicalpsychomotor reactivity to amphetamine (Anagnostaras et al.2000; Yates et al. 2007). Finally, B6 and H mice are themost common background strains in studies using targetedmutations, and the present studies are part of a larger planof study that involves several mutants (Crawley et al.1997; Silva et al. 1997). H mice were used in experiment 2due to availability but perform comparably on this task(Anagnostaras et al. 2003).

Apparatus

Conditioning context Testing was performed in a window-less room. Background noise (65 dB) was provided by aHEPA air cleaner, and white light was provided by two100-W bulbs. The mice were continuously recorded by awall-mounted color video camera that was connected to acomputer and video equipment in an adjacent room. Threeto four mice were tested concurrently in individualconditioning chambers. Each chamber (32 cm wide,25 cm high, 25 cm deep) was equipped with a stainless

198 Psychopharmacology (2009) 202:197–206

steel grid floor (36 rods, each rod 2 mm in diameter, 8 mmcenter to center) and a speaker in the side wall. The sidewalls were white acrylic, and the front wall was clear toallow for viewing (Med-Associates Inc., St. Albans, VT,USA). A stainless steel drop-pan, scented with 7%isopropyl alcohol to provide a background odor, waslocated beneath each chamber. Between tests, the condi-tioning contexts were cleaned with 7% isopropyl alcoholsolution. Each chamber was connected to a solid-statescrambler, providing AC constant current shock, and anaudio stimulus generator was located in an adjacent room,controlled via an interface connected to a Windowscomputer running Med-PC (Med-Associates Inc., StAlbans, VT, USA). Automated assessment of freezing andactivity was provided by custom-designed software adaptedfrom NIH Image running on an Apple Macintosh G4(automated algorithm validated elsewhere; Anagnostaraset al. 2000).

Alternate context Multiple (three to four) mice were testedconcurrently for tone fear in a separate room in individualboxes measuring 30 cm wide, 25 cm high, and 24 cm deep,and equipped with a speaker in the side walls. A clearPlexiglas front wall allowed the mice to be continuallyobserved, while the ceiling, floor, and three interior walls ofthe chamber were solid white. To create a space distinctfrom the training context, a white plastic, triangular tentwas placed inside each box; each side of the trianglemeasured 23 cm. Between tests, the chambers were cleanedand scented with a 5% white vinegar solution. The roomwas lit with dim red light, and an infrared video camera,connected to the Macintosh G4 described above, was usedto score freezing.

Drugs Drugs were administered intraperitoneally (i.p.) in avolume of 5 or 10 ml/kg. D-Amphetamine sulfate (Sigma-Aldrich Co., St. Louis, MO, USA) was dissolved in 0.9%physiological saline. Amphetamine injections (salt weight;0.005, 0.025, 0.05, 0.5, 1, 2, 4, or 8 mg/kg) were given15 min before introduction to the testing equipment.

Experiment 1: behavior measurement In order to measurethe effects of amphetamine on exploratory locomotoractivity, baseline activity in 2 min before the first tone–shock pairing on the training day was assessed by countingthe number of cross-overs each subject performed (Marenet al. 1998). A single cross-over was defined as themovement of a subject’s entire body from one half of thebox to the other. Videotapes of the conditioning sessionswere observed using a standard VCR and monitor. Inaddition to number of cross-overs, exploratory activity wasalso measured using an automated, video-based activitymeasure (Anagnostaras et al. 2000).

To assess whether amphetamine disrupted shock reactivity,mouse activity burst displayed during the 2 s of shockexposure (unconditioned response to shock), as well asactivity during the 2 s leading up to the shock, wasmeasured as velocity (centimeter per second; Anagnostaraset al. 2000).

Experimental procedures

Experiment 1

Conditioning Mice were injected with saline or amphet-amine 15 min prior to training. Training consisted of a 2-minbaseline period, followed by three tone–shock pairings,each separated by 30 s. A tone–shock pairing consisted of a30-s tone (2.8 kHz, 85 dB, A Scale), with a scrambled,constant current AC footshock (0.75 mA) administeredduring the last 2 s of the tone. Immediate post-shockfreezing was measured for another 5 min. Thus, mice wereinside the fear-conditioning chambers for a total of 10 minbefore being returned to their home cages.

Testing Mice were returned to the conditioning contextwithout drug 24 h after training, and freezing was scoredfor 5 min. Mice were placed in the alternate context 48 hafter training, also off-drug. Tone testing consisted of a2-min baseline, followed by a continuous 3-min toneidentical to the training tone. Freezing was scored for theentire 5-min period.

Experiment 2

Procedures for experiment 2 were identical to experiment 1,with the exception that amphetamine or saline injectionswere administered 15 min before testing, in addition tothose administered before training.

Results

Experiment 1

Generalized activity Amphetamine produced a dose-dependent increase in activity during the 2-min baselineperiod of training, measured both by an automated computerscoring system and by hand-counted cage cross-overs(Fig. 1a). Group differences in cage cross-overs were foundon an analysis of variance [ANOVA; F(8, 127)=16.5, p<0.0001]. Subjects administered with 4–8 mg/kg amphet-amine pre-training displayed significantly more cross-oversduring baseline than the saline control group [Fisher’sprotected least significant difference (PLSD) multiple post

Psychopharmacology (2009) 202:197–206 199

hoc comparisons, p values <0.01]. No other doses differedsignificantly from saline controls (p values >0.05). AnANOVA demonstrated differences in the automated activitymeasure as well [F(8, 127)=7.5, p<0.0001]. As with thecross-over measures, only mice given 4–8 mg/kg ofamphetamine showed a significant difference in activityfrom saline controls (Fisher’s PLSD, p values <0.01).

Activity burst velocity A large difference in velocity waselicited by the first 2-s shock presentation [unconditionedresponse (UR)], compared with the preceding 2 s, duringwhich no shock was present (Fig. 1b). A multivariateANOVA revealed a significant effect of dose on velocity [F(8, 127)=2.3, p<0.05], a significant effect of the shock [URversus baseline; F(1, 127)=1687.5, p<0.0001], along witha significant time period by dose interaction [F(8, 127)=2.5, p<0.05], so the baseline and UR were considered

separately. There were significant differences in thebaseline velocity [F(8, 127)=5.8, p<0.0001]. Post hoccomparisons revealed similar differences in velocity asthose seen in measurements of baseline activity, withsubjects administered with 1, 4, or 8 mg/kg amphetaminedisplaying increased velocity compared with subjectsadministered with saline (Fisher’s PLSD, p values <0.05).No other doses differed significantly in terms of baselineactivity from saline. There were also significant differencesduring the UR [F(8, 127)=2.2, p<0.05]. Only subjectsgiven 0.025 or 8 mg/kg exhibited a larger UR to the shockcompared to controls (p values <0.05), but these differenceswere small and unrelated to the memory effects. Thus,amphetamine did not dull shock reactivity and seemed toenhance reactivity in certain cases.

Training Subjects were inside the conditioning contexts onthe training day for a total of 10 min. Figure 2a depicts thefirst 5 min of training, consisting of a 2-min baselineperiod, followed by three tone–shock pairings during3 min. A main effect for dose was present within the5-min period [F(8, 127)=36.5, p<0.0001], as was a maineffect for minute [F(4, 127)=248.1, p<0.0001], along witha significant dose by minute interaction [F(32, 127)=15.4,p<0.0001]. Post hoc comparisons revealed a dose-depen-dent effect of amphetamine on freezing, with subjects thathad received 1, 2, 4, or 8 mg/kg freezing less than salinecontrols (Fisher’s PLSDs, p values <0.0001). Interestingly,subjects that had been administered with 0.005, 0.025, or0.05 mg/kg amphetamine froze more than the saline controlsubjects (p values <0.01). Subjects that received 0.5 mg/kgwere not significantly different from saline controls (p>0.05).

Post-shock freezing In minutes 6 through 10 of training,freezing was measured with no further presentation of toneor shock, as an index of post-shock freezing (Fig. 2b).Dose-dependent differences in freezing were apparent [F(8,127)=68.7, p<0.0001]. Post hoc tests revealed that subjectsthat had received 1, 2, 4, or 8 mg/kg amphetaminedisplayed less freezing than saline controls (Fisher’s PLSD,p values <0.001). As these subjects were still on-drug, it isimportant to note that amphetamine’s locomotor effectscould be influencing these results. Interestingly, however,subjects administered with 0.005, 0.025, or 0.05 pre-training again froze more than saline controls (p values<0.05). Subjects that received 0.5 mg/kg were not signif-icantly different from saline controls (p>0.05).

Context fear Subjects were returned to the conditioningchambers 24 h after training. Freezing was measured for5 min, with all subjects off-drug (Fig. 3a). Dose-dependentdifferences in freezing were apparent [F(8, 127)=20.7, p<0.0001]. Subjects that had previously received 4 or 8 mg/kg

0

5

10

15

20

25

30

00.05

0.5 1 2 4 8

Dose (mg/kg)0.005

0.025

0 0.5 1 2 4 80.005

0.0250.05

Baseline ActivityA

Crossovers

Video (au)

Baseline

00.05

0.5 1 2 4 8

Dose (mg/kg)

0.0050.025

0

Vel

oci

ty (

cm/s

)

5

10

15

20

25

30

35

Shock

Shock ReactivityB

*

*

**

Fig. 1 a Baseline activity. A dose-dependent increase in activity wasobserved during the 2-min baseline period prior to the first tone–shockpairing, with subjects administered with 4 or 8 mg/kg amphetamineprior to training showing greater activity than saline controls (left, anautomated video activity measure and right, full cage cross-overs;mean ± SEM). Asterisks indicate a significant difference from salinecontrols. b Shock reactivity. All groups showed a significant increasein velocity (centimeter per second, mean ± SEM) during the 2-s shockcompared with the 2-s baseline period leading up to the shock

200 Psychopharmacology (2009) 202:197–206

amphetamine before training displayed less freezing off-drug when reintroduced to the training context comparedwith saline controls (Fisher’s PLSD, p values <0.01). Noother groups differed from saline controls (p values >0.05).

Tone fear Subjects were introduced to a new context 48 hafter training off-drug. Freezing was measured for 5 min,consisting of a 2-min baseline period and a 3-min tonepresentation (Fig. 3b). The tone was the same frequencyand volume as that which had been paired with the shockduring training. Dose-dependent differences in freezingduring the tone presentation were apparent [F(8, 127)=17.1, p<0.0001]. Only subjects that had received 8 mg/kgamphetamine pre-training showed decreased freezing dur-ing the tone test compared with saline controls (Fisher’sPLSD, p<0.0001). Those that had been injected with 0.005,

0.025, or 0.05 mg/kg amphetamine pre-training againshowed increased freezing compared with saline controls(p values <0.001). No other doses differed from salinecontrols (p values >0.05).

Experiment 1: summary Overall, contextual fear memorytested off-drug was sensitive to disruption by 4–8 mg/kg ofamphetamine, whereas tone fear was sensitive only to8 mg/kg. An enhancement in cued, but not contextual,memory was observed at 0.005, 0.025, and 0.05 mg/kg.

Contextual Fear (Off-Drug)a

0

10

Fre

ezin

g (

%)

Fre

ezin

g (

%)

20

30

40

50

60

70

80

00.05

0.5 1 2 4 8

Dose (mg/kg)0.005

0.025

Tone Fear

(Off-Drug)

00.05

0.5 1 2 4 8

Dose (mg/kg)0.005

0.0250

10

20

30

40

50

60

70

80b

*

*

*

* **

Fig. 3 a Contextual fear. When returned to the training context 24 hafter training off-drug, subjects that had received 4 or 8 mg/kg pre-training amphetamine displayed less freezing (mean ± SEM) than allother groups. b Tone fear. Freezing (mean ± SEM) during the 3-minperiod during the test when the tone was presented is depicted. Thosethat had been injected with 8 mg/kg pre-training displayed lessfreezing than all other groups, while those that had been administeredwith 0.005, 0.025, or 0.05 mg/kg displayed more freezing than allother groups

Minute1 2 3 4 5

0

10

Fre

ezin

g (

%)

Fre

ezin

g (

%)

20

30

40

50

60

70

0.05

0

0.5

1

2

8

4

mg/kg

0.005

0.025

Training(On-Drug)

a

0

10

20

30

40

50

60

70

80

90

b Post-Shock Freezing(On-Drug)

00.05

0.5 1 2 4 8

Dose (mg/kg)

0.0050.025

* * *

*

*

**

**

*

****

Fig. 2 a Training. Freezing (mean ± SEM) of subjects on 1, 2, 4, or8 mg/kg amphetamine was less than saline control subjects, whilefreezing of subjects on 0.005, 0.025, or 0.05 mg/kg amphetamine wasgreater than controls. b Post-shock freezing. Dose-dependent, post-training freezing was observed, with subjects on 0.005, 0.025, or0.05 mg/kg amphetamine freezing more than controls, while those on1, 2, 4, or 8 mg/kg amphetamine displayed less freezing than controls

Psychopharmacology (2009) 202:197–206 201

Experiment 2

Training and post-shock freezing The training protocol forthe state-dependent control study was the same as the dose–response study. Fifteen minutes after drug or salineinjection, subjects were exposed to the conditioningchambers for a total of 10 min, composed of a 2-minbaseline period, 3 min during which three tone–shockpairings were presented, and a 5-min immediate post-shockfreezing test (Fig. 4a). Drug-dependent effects on freezingwere evident [F(3, 36)=88.7, p<0.0001], with miceinjected with saline pre-training showing markedly morefreezing throughout training than those injected with 8 mg/kgamphetamine (Fisher’s PLSD, p values <0.0001). A maineffect for training period was evident as well [F(2, 36)=259.9, p<0.0001], with average freezing increasingthroughout training (p values <0.0001). A significant drugby training period interaction [F(6, 36)=59.7, p<0.0001],however, demonstrated that this effect was driven differen-tially by the two saline groups, as the two groups onamphetamine did not show an increase in freezing.

Context fear Using the same protocol as experiment 1,subjects were returned to the conditioning chambers 24 hafter training to measure contextual fear memory (Fig. 4b).Drug treatment had significant effects on contextual fearexpression [F(3, 36)=66.5, p<0.0001], with both groupsinjected with 8 mg/kg amphetamine pre-testing displayingless freezing than those injected with saline (Fisher’s PLSD,p values <0.0001). Mice that received amphetamine duringtesting exhibited very little freezing and did not differ fromone another (p>0.05). Consistent with experiment 1, micethat had been injected with 8 mg/kg amphetamine pre-training but then injected with saline pre-testing displayedless contextual freezing than those that had been injectedwith saline both pre-training and pre-testing (p<0.0001).Overall, no state-dependent learning effects were evident.Consistent with experiment 1, amphetamine was able todisrupt performance of freezing when animals were trainedon-drug, as well as contextual freezing when those animalswere tested 24 h later off-drug.

Discussion

We predicted that amphetamine would disrupt Pavlovianfear conditioning at moderate to high doses, similar to thosethat addicts take, and enhance fear conditioning at lowdoses, similar to those used to treat ADHD. As expected,moderate doses of amphetamine administered pre-traininginhibited contextual and cue-elicited freezing off-drug,while very low doses enhanced cued fear memory. These

data are in agreement with the results of our previous workwith cocaine, which induced a similar pattern of enhance-ments and deficits in fear conditioning (Wood et al. 2007).A low dose of cocaine (0.1 mg/kg) produced an enhance-ment in contextual and cued fear, whereas a high dose

a

0

10

20

30

40

50

60

70

80

100

90

Training & Post-Shock Freezing(State-Dependent Control)

b

0

10

20

30

40

50

60

Fre

ezin

g (

%)

Fre

ezin

g (

%)

70

80

90

TestingTreatment

Contextual Fear(State-Dependent Control)

Training AMPHSalineAMPH

AMPH SalineAMPH

SalineSaline

1-2 3-5 5-10

SS

SA

AS

AA

Minute

*

Fig. 4 a State-dependent control—training and post-shock freezing.While subjects administered with saline before training showed asteady increase in freezing during training, subjects administered with8 mg/kg amphetamine showed no increase in freezing. b State-dependent control—contextual fear. Regardless of treatment beforetraining, subjects administered with 8 mg/kg amphetamine beforecontextual testing displayed very little freezing. In contrast, miceinjected with saline before contextual testing showed some freezingif injected with 8 mg/kg before training and greater levels of freezingif injected with saline before training (asterisk indicates significantdifference between these two groups). There was no evidence of state-dependent memory

202 Psychopharmacology (2009) 202:197–206

(15 mg/kg) produced an impairment in both measures. Thecombined results of these studies lead us to theorize that itis perhaps not the addictive properties of the drugs or theclinical profile of the subjects taking the drugs but, rather,the dose of drugs that lead to the divergent cognitive effectsassociated with psychostimulants.

One difference in the results of the present study and ourprevious research with cocaine (Wood et al. 2007) is that arange of low doses of amphetamine studied did not enhancecontextual fear 24 h after training. It is unclear if thememory enhancement provided by a low dose of amphet-amine is specific to cued fear as immediate post-shockfreezing is believed to be a form of contextual fear(Fanselow 1986). Indeed, while low doses of amphetamineinduced small but significantly higher levels of freezingduring the immediate post-shock freezing test compared tosaline controls, levels of freezing off-drug during the tonetest were almost double that of controls. This evidencesuggests the effects on locomotor activity and freezing aredissociable, especially for the low doses. It may be the casethat some other, untested low dose of amphetamine wouldreveal enhanced memory on both contextual and cued testsoff-drug, but it seems likely that a very low dose ofamphetamine enhances cued memory, exclusively, since wetried a wide range of low doses (0.005–0.5 mg/kg) andinduced large enhancements of tone memory. Anotherpossibility is that the enhancement was not observed incontextual fear because of a “ceiling effect”; about 70%freezing is potentially the maximum that B6 mice mayfreeze under these training parameters (Anagnostaras et al.2000, 2003). Clearly, a moderate or high dose of amphet-amine disrupts conditioned freezing; there were somedifferences in terms of the deficits, in that 4 and 8 mg/kgproduced significant deficits in contextual freezing, whereonly 8 mg/kg was effective in disrupting tone freezing.However, overall, the deficits in context and tone condi-tioning seem similar at high doses.

Research with a variety of human populations hasproduced mixed findings about the cognitive effects ofstimulant use in humans. One study found cognitivebenefits with amphetamine administration (0.25 mg/kg,orally) to people with schizophrenia already treated withantipsychotics (Barch and Carter 2005) despite the commonnotion that amphetamine exacerbates schizophrenia. Thenon-schizophrenic control subjects of the study, as well,exhibited improved language production and reaction timebut had no benefit in working memory accuracy. Inaddition, it has been reported that low-dose amphetamine(10 mg, p.o., D-amphetamine, approximately 0.12 mg/kgfor an 80-kg person) given during speech therapy leads toimproved recovery from aphasia (Walker-Batson et al.2001). These findings demonstrate the ability of stimulantsto improve certain cognitive functions in a healthy

population. Many studies have found cognitive deficits inamphetamine and cocaine addicts (Rogers and Robbins2001). Currently using methamphetamine abusers werefound to have worse performance on word recall, percep-tual speed, an abbreviated IQ test, and the Wisconsin CardSorting Task, compared with controls (Simon et al. 2002).In addition, methamphetamine addicts were found toperform more poorly on a decision-making task than theirmatched controls while displaying less activation in anumber of frontal cortical areas, as measured by functionalmagnetic resonance imaging (fMRI; Paulus et al. 2002).

Previous research on amphetamine administration andassociative learning in rodents has also yielded mixedfindings. Rats administered with D-amphetamine (5 mg/kg,i.p.) or cocaine (40 mg/kg, i.p.) during extinction of fear-potentiated startle continued to show high startle amplitudesafter 120 presentations of the nonreinforced CS (Borowskiand Kokkinidis 1998). These results indicate that amphet-amine and cocaine impaired the extinction of fear-potentiatedstartle. Extinction is generally conceived as the acquisition ofnew, inhibitory learning (Barad 2006). These doses ofamphetamine and cocaine would produce impairments ifgiven during training on our conditioned freezing task. Post-training injections of D-amphetamine at 1.0 mg/kg (i.p.), butnot 0.25 or 4.0 mg/kg, have been shown to enhanceinhibitory avoidance learning (Martinez et al. 1980). It isunclear if pre- versus post-training injections would share thesame dose–response curve, so it is difficult to directlycompare the results of the Martinez study with the currentresults. In addition, in order to examine amphetamine’seffects on experience-dependent plasticity in rats, researchersexamined fear conditioning in rats housed in a complexenvironment for 3 months (Briand et al. 2005). While ratsshowed enhanced learning after living in the complexenvironment, rats treated with D-amphetamine (4.0 mg/kg)for 21 days, then housed in the enriched environment for3 months, did not show the same enhancement. The authorshypothesized that this moderate dose of amphetamine limitedthe experience-dependent structural plasticity normallyengaged by a complex environment.

As with cocaine, several possible reasons exist as to whyconditioned freezing was disrupted at high doses (Wood etal. 2007). While amphetamine does not have anestheticproperties, it could have induced a positive hedonic state inthe mice. Those on higher doses may have perceived theshocks as less aversive than those on lower doses becauseof a drug-induced feeling of well being. While thisexperiment is not designed to fully explore this possibility,there is little evidence that this explanation accounts for theresults. We measured the shock reactivity of subjects(Fig. 1b) and found that all subjects showed a largeincrease in velocity during the first shock presentation, incomparison to baseline velocity. Mice on 8 mg/kg

Psychopharmacology (2009) 202:197–206 203

amphetamine seemed to perceive the shock at least asaversive as those on lower doses. In addition, thisexplanation of altered perception would imply that amphet-amine produces a positive hedonic state disruptive to fearconditioning, whereas much research has shown thatamphetamine is anxiogenic, enhancing most defensivebehaviors in rodents (Markham et al. 2006).

Previous research on a range of doses of amphetaminesupports an interpretation of attentional deficits disruptingperformance. Rats showed impairment on a delayedmatching-to-sample task at low doses of 0.6 and 1.0 mg/kgamphetamine, i.p. (Harper et al. 2005). At these doses,considered low doses for the present study on mice, theresponses for a given trial were greatly influenced byresponses from the preceding trial. These results wereinterpreted to stem from poor attention or confusion, asopposed to a pure associative memory deficit. Evidence forattentional deficits due to chronic amphetamine (5.0 mg/kg,i.p.) use was also found in extinction deficits of active andpassive avoidance responses in mice (Kokkinidis 1983).Human studies, utilizing techniques such as fMRI (Paulus etal. 2002), have revealed frontal lobe dysfunction inamphetamine-dependent individuals, resulting in poor atten-tion and decision making. Taken together, there is evidencefrom previous work that amphetamine could be hinderingperformance on fear conditioning due to attentional ratherthan learning deficits.

An additional, potential confound to be addressed isstate-dependent learning. According to this theory, subjectson-drug during training would remember what they learnedbetter if also tested on-drug. A discrepancy in drug statebetween training and testing has been shown to bedetrimental to learning, for certain drugs. For example,one representative study examined the effects of benzodi-azepine [chlordiazepoxide (CDP)] administration duringextinction of fear conditioning (Bouton et al. 1990). Theyfound that extinction was state dependent when done onCDP. However, in the present study using amphetamine, wefound no evidence of state-dependent learning (Fig. 4). Inthe present study, subjects that received 0.005, 0.025, or0.05 mg/kg amphetamine pre-training displayed higherlevels of freezing than controls when tested off-drug. Thisfinding contradicts the prediction of state-dependent learn-ing. In addition, the immediate post-shock freezing test,performed on-drug, elicited a greater deficit in freezing thanwhen subjects were tested off-drug. While an immediatememory test for a study of this design is clearly confoundedwith the motor stimulant properties of amphetamine, it isworthy to note that animals administered with a high doseof amphetamine during both testing and training showed noincrease in freezing over those that had received drug onlyduring testing in experiment 2. Indeed, subjects givenamphetamine during both training and testing exhibited

almost no evidence of learning, suggesting that there wasno state-dependent learning. Rather, amphetamine appearsto have disrupted performance of the freezing responsewhen given during testing and not training and disruptedthe acquisition of fear conditioning when given duringtraining and not testing. Interpretation of this data, thus, issomewhat problematic since high-dose amphetamine mayhave had two separate actions: disrupting memory forma-tion and/or disrupting freezing by producing hyperactivity.Overall, however, amphetamine and other stimulants havenot been shown to produce a high level of state-dependentlearning compared with other addictive drugs (Overton1972).

Studies on the effects of amphetamine on long-termpotentiation (LTP), an experience-dependent increase insynaptic efficacy thought to be associated with learning,have produced conflicting results. Broadly, some researchon methamphetamine users has shown a decreased densityof dopamine transporter in the striatum (Lundqvist 2005).As activation of DA receptors can modulate the expressionof glutamate receptors, DA receptor activity can, in turn,alter synaptic plasticity (Sun et al. 2005). LTP in the dentategyrus was also found to be enhanced by amphetamine in adose-dependent manner (Gold et al. 1984). Increasedpotentiation of population spike amplitude was found inrats that had been administered with 0.01, 0.1, 1.0, or3.0 mg/kg amphetamine, i.p., while doses higher (10.0 mg/kg)and lower (0.001 mg/kg) produced no differences com-pared to controls. In other research, however, acuteamphetamine administration (5.0 mg/kg, i.p.) did not alterperforant path LTP of EPSP slope but caused a smallreduction in LTP of the population spike amplitude(Morimoto et al. 1987). The authors interpreted theseresults as showing amphetamine reduces cellular excit-ability. LTP was blocked in nucleus accumbens neuronswhen slices were bathed in 2.5 μM amphetamine solution(Li and Kauer 2004). Interestingly, this effect wasattenuated when rats were administered with amphetamine(2.5 mg/kg, i.p.) in vivo for 6 days, and nucleusaccumbens slices were prepared 8–10 days after the lastinjection. Additionally, a higher AMPA-receptor/NMDA-receptor ratio at the glutamatergic synapses in the ventraltegmental area (VTA) was associated with higher behav-ioral sensitization to a single dose of amphetamine inyoung rats (Faleiro et al. 2004). These results indicate thatLTP occurred rapidly in the VTA and with a singleamphetamine exposure. Overall, studies of amphetamineand synaptic plasticity have yielded mixed results butdemonstrate that low doses of amphetamine appear toenhance cell excitability and synaptic plasticity.

In general, research in humans and rodents on memoryand synaptic plasticity is in agreement with the currentstudy for acute use of amphetamine, suggesting that very

204 Psychopharmacology (2009) 202:197–206

low doses are beneficial to cognition while moderate orhigh doses are detrimental. We have previously foundsimilar results with cocaine (Wood et al. 2007). Theimplications of this research are broad, ranging fromaddicts’ use to students using stimulants for academicdoping and those prescribed stimulants for learningdisabilities. An enhanced understanding of when thesedrugs produce improvements in cognition, as opposed todeficits and addiction, is needed. The present studysuggests that dosing is a crucial determinant of stimulanteffects on cognition. At very low doses, cognition isenhanced, whereas high doses promote addiction anddisrupt cognition. This hypothesis seems more parsimoni-ous, given the sum of the data, than other theories that positthat stimulants act differently in children versus adults or insubjects with or without ADHD (Marx 1999).

Acknowledgments We thank Jakyong Lee, Angela Smith, andJason Yeatman for excellent technical assistance; we also thankTristan Shuman and Jennifer Sage for helpful comments on themanuscript. SCW was supported by an NSF Graduate ResearchFellowship. These studies were supported by NIH grant (DA020041)and Hellman Fellowship to SGA. The above experiments comply withthe current laws of the United States, the country in which they wereperformed.

References

Ahmann PA, Theye FW, Berg R, Linquist AJ, Van Erem AJ, CampbellLR (2001) Placebo-controlled evaluation of amphetamine mixture-dextroamphetamine salts and amphetamine salts (Adderall):efficacy rate and side effects. Pediatrics 107:E10

Anagnostaras SG, Maren S, Fanselow MS (1999) Temporally gradedretrograde amnesia of contextual fear after hippocampal damagein rats: within-subjects examination. J Neurosci 19:1106–1114

Anagnostaras SG, Josselyn SA, Frankland PW, Silva AJ (2000)Computer-assisted behavioral assessment of Pavlovian fearconditioning in mice. Learn Mem 7:58–72

Anagnostaras SG, Gale GD, Fanselow MS (2001) Hippocampus andcontextual fear conditioning: recent controversies and advances.Hippocampus 11:8–17

Anagnostaras SG, Murphy GG, Hamilton SE, Mitchell SL, RahnamaNP, Nathanson NM, Silva AJ (2003) Selective cognitivedysfunction in acetylcholine M1 muscarinic receptor mutantmice. Nat Neurosci 6:51–58

Barad M (2006) Is extinction of fear erasure or inhibition? Why both,of course. Learn Mem 13:108–109

Barch DM, Carter CS (2005) Amphetamine improves cognitivefunction in medicated individuals with schizophrenia and inhealthy volunteers. Schizophr Res 77:43–58

Borowski TB, Kokkinidis L (1998) The effects of cocaine, amphet-amine, and the dopamine D1 receptor agonist SKF 38393 on fearextinction as measured with potentiated startle: implications forpsychomotor stimulant psychosis. Behav Neurosci 112:952–965

Bouton ME, Kenney FA, Rosengard C (1990) State-dependent fearextinction with two benzodiazepine tranquilizers. Behav Neurosci104:44–55

Briand LA, Robinson TE, Maren S (2005) Enhancement of auditory fearconditioning after housing in a complex environment is attenuatedby prior treatment with amphetamine. Learn Mem 12:553–556

Butcher J (2003) Cognitive enhancement raises ethical concerns.Academics urge pre-emptive debate on neurotechnologies.Lancet 362:132–133

Caldwell JA, Caldwell JL, Crowley JS, Jones HD (1995) Sustaininghelicopter pilot performance with Dexedrine during periods ofsleep deprivation. Aviat Space Environ Med 66:930–937

Cornum K, Cornum R, Storm W (1995) Use of psychostimulants inextended flight operations: a Desert Shield experience AerospaceMedical Panel Symposium (Advisory Group for AerospaceResearch and Development Conference Proceedings 579,Neurological Limitations of Aircraft Operations: HumanPerformance Implications). Koln, Germany, pp 371–374

Cornum KG (1994) Extended air combat operations: F-15s over Iraq.Aerospace Medical Association 65th Annual Scientific Meeting.Aerospace Medical Association, Washington, p A49

Crawley JN, Belknap JK, Collins A, Crabbe JC, Frankel W, HendersonN, Hitzemann RJ, Maxson SC, Miner LL, Silva AJ, Wehner JM,Wynshaw-Boris A, Paylor R (1997) Behavioral phenotypes ofinbred mouse strains: implications and recommendations formolecular studies. Psychopharmacology (Berl) 132:107–124

Ellinwood EH, King GR, Lee TH (1998) Chronic amphetamine useand abuse. In: Bloom FE, Kupfer DJ (eds) Psychopharmacology:the fourth generation of progress. CD Rom Version. 4th edn.Lippincott, Williams and Wilkins, Philadelphia

Faleiro LJ, Jones S, Kauer JA (2004) Rapid synaptic plasticity ofglutamatergic synapses on dopamine neurons in the ventraltegmental area in response to acute amphetamine injection.Neuropsychopharmacology 29:2115–2125

Fanselow MS (1980) Conditioned and unconditional components ofpost-shock freezing. Pavlov J Biol Sci 15:177–182

Fanselow MS (1986) Associative vs. topographical accounts of theimmediate shock freezing deficit in rats: implications for theresponse selection rules governing species specific defensivereactions. Learning & Motivation 17:16–39

Fanselow MS, Gale GD (2003) The amygdala, fear, and memory. AnnNY Acad Sci 985:125–134

Fanselow MS, Poulos AM (2005) The neuroscience of mammalianassociative learning. Annu Rev Psychol 56:207–234

Gold PE, Delanoy RL, Merrin J (1984) Modulation of long-termpotentiation by peripherally administered amphetamine andepinephrine. Brain Res 305:103–107

Harper DN, Wisnewski R, Hunt M, Schenk S (2005) (+/−)3,4-methylenedioxymethamphetamine, D-amphetamine, and cocaineimpair delayed matching-to-sample performance by an increasein susceptibility to proactive interference. Behav Neurosci119:455–463

Harris D, Batki SL (2000) Stimulant psychosis: symptom profile andacute clinical course. Am J Addict 9:28–37

Kessal K, Chessel A, Spennato G, Garcia R (2005) Ketamine andamphetamine both enhance synaptic transmission in the amyg-dala-nucleus accumbens pathway but with different time-courses.Synapse 57:61–65

Kokkinidis L (1983) The effects of chronic amphetamine administrationon the acquisition and extinction of an active and passive avoidanceresponse in mice. Pharmacol Biochem Behav 19:593–598

Leonard BE, McCartan D, White J, King DJ (2004) Methylphenidate:a review of its neuropharmacological, neuropsychological andadverse clinical effects. Hum Psychopharmacol 19:151–180

Li Y, Kauer JA (2004) Repeated exposure to amphetamine disruptsdopaminergic modulation of excitatory synaptic plasticity andneurotransmission in nucleus accumbens. Synapse 51:1–10

Lundqvist T (2005) Cognitive consequences of cannabis use:comparison with abuse of stimulants and heroin with regard to

Psychopharmacology (2009) 202:197–206 205

attention, memory and executive functions. Pharmacol BiochemBehav 81:319–330

Maren S, Anagnostaras S, Fanselow M (1998) The startled seahorse:is the hippocampus necessary for contextual fear conditioning?Trends Cogn Sci 2:39–42

Markham CM, Yang M, Blanchard RJ, Blanchard DC (2006) Effectsof D-amphetamine on defensive behaviors related to fear andanxiety. Pharmacol Biochem Behav 83:490–499

Martinez JL Jr, Jensen RA, Messing RB, Vasquez BJ, Soumireu-Mourat B, Geddes D, Liang KC, McGaugh JL (1980) Centraland peripheral actions of amphetamine on memory storage. BrainRes 182:157–166

Marx J (1999) How stimulant drugs may calm hyperactivity. Science283:306

McCabe SE, Knight JR, Teter CJ, Wechsler H (2005) Non-medicaluse of prescription stimulants among US college students:prevalence and correlates from a national survey. Addiction100:96–106

Morimoto K, Otani S, Goddard GV (1987) Effects of acute and long-termtreatment with amphetamine on evoked responses and long-termpotentiation in the dentate gyrus of anesthetized rats. Brain Res407:137–143

Overton DA (1972) State-dependent learning produced by addictingdrugs. In: Fisher C, Freedman AM (eds) Opiate addiction: originsand treatment. Halsted Press Division of Wiley, New York, pp61–75

Paulus MP, Hozack NE, Zauscher BE, Frank L, Brown GG, Braff DL,Schuckit MA (2002) Behavioral and functional neuroimagingevidence for prefrontal dysfunction in methamphetamine-dependentsubjects. Neuropsychopharmacology 26:53–63

Rogers RD, Robbins TW (2001) Investigating the neurocognitivedeficits associated with chronic drug misuse. Curr Opin Neurobiol11:250–257

Silva AJ, Simpson EM, Takahashi JS, Lipp H, Nakanishi S, Wehner JM,Giese KP, Tully T, Abel T, Chapman PF, Fox K, Grant S, Itohara S,Lathe R,MayfordM,McNamara JO,Morris RJ, Picciotto M, RoderM, Shin H, Slesinger PA, Storm DR, Stryker MP, Tonegawa S,Wang Y, Wolfer DP (1997) Mutant mice and neuroscience:recommendations concerning genetic background. BanburyConference on genetic background in mice. Neuron 19:755–759

Simon SL, Domier CP, Sim T, Richardson K, Rawson RA, Ling W(2002) Cognitive performance of current methamphetamine andcocaine abusers. J Addict Dis 21:61–74

Sun X, Zhao Y, Wolf ME (2005) Dopamine receptor stimulationmodulates AMPA receptor synaptic insertion in prefrontal cortexneurons. J Neurosci 25:7342–7351

Teter CJ, McCabe SE, LaGrange K, Cranford JA, Boyd CJ (2006)Illicit use of specific prescription stimulants among collegestudents: prevalence, motives, and routes of administration.Pharmacotherapy 26:1501–1510

Turner DC, Robbins TW, Clark L, Aron AR, Dowson J, Sahakian BJ(2003) Cognitive enhancing effects of modafinil in healthyvolunteers. Psychopharmacology (Berl) 165:260–269

Walker-Batson D, Curtis S, Natarajan R, Ford J, Dronkers N,Salmeron E, Lai J, Unwin DH (2001) A double-blind, placebo-controlled study of the use of amphetamine in the treatment ofaphasia. Stroke 32:2093–2098

White IM, Whitaker C, White W (2006) Amphetamine-inducedhyperlocomotion in rats: hippocampal modulation of the nucleusaccumbens. Hippocampus 16:596–603

Wood SC, Fay J, Sage JR, Anagnostaras SG (2007) Cocaine andPavlovian fear conditioning: dose–effect analysis. Behav BrainRes 176:244–250

Yates JW, Meij JT, Sullivan JR, Richtand NM, Yu L (2007) Bimodaleffect of amphetamine on motor behaviors in C57BL/6 mice.Neurosci Lett 427:66–70

206 Psychopharmacology (2009) 202:197–206